Nvidia eyes Mitsubishi Heavy cooling tech for AI data centers

Nvidia ↘ 0.92% is considering adopting Mitsubishi Heavy Industries' cooling systems and energy management technology for its next-generation AI data center facilities, according to a report by Nikkei on Monday.streetinsider+2

The Potential Partnership

The discussions would see Mitsubishi Heavy supply critical infrastructure components — including cooling systems, emergency power, and energy management equipment — to support Nvidia's expanding network of AI data centers. The talks reflect growing demand for advanced thermal and power solutions as AI chips push data center energy density to new levels.fidelity+1

MHI's Data Center Ambitions

The potential deal comes as MHI has been aggressively positioning itself as a one-stop provider for data center infrastructure. The Japanese industrial conglomerate established a strategic business base in Dallas, Texas, in May 2025 to strengthen its North American data center operations. At Nvidia's GTC conference in San Jose in March 2026, MHI showcased high-efficiency cooling solutions and on-site power generation technology.mhi+3

MHI has developed expertise in two-phase direct-to-chip cooling systems, which it began deploying commercially in late 2025 in partnership with EXEO Group. The company also launched DIAVAULT, an industrial-grade edge data center platform integrating power and cooling systems, in February 2026.acnnewswire+1

A Rapidly Growing Market

The data center liquid cooling market is projected to expand from $6.6 billion in 2026 to $38.4 billion by 2033, driven largely by the thermal demands of AI training and inference workloads. Nvidia's Blackwell-generation racks have pushed thermal loads to levels requiring sophisticated cooling infrastructure beyond traditional air-based approaches.channelchek+1

A formal partnership with Nvidia would represent a major validation of MHI's integrated approach to data center infrastructure, combining its industrial heritage in power generation and thermal systems with the specific demands of AI computing at scale.